Review

Invited
Dynamic Image Interpretation of MRI of the Breast
Christiane K. Kuhl, MD
*
and Hans H. Schild, MD
Dynamic breast MRI provides information on both lesion
cross-sectional morphology and functional lesion features
such as vascularity/perfusion and vessel permeability. This
review gives an overview of the historical background of dy-
namic contrast-enhanced breast MRI. It explains the tech-
niques pathophysiological basis, describes the various
technical approaches that have been pursued and the corre-
sponding interpretation guidelines that have been proposed
(including their respective diagnostic accuracies), and pre-
sents established and evolving clinical applications of the
dynamic approach to breast MRI. J. Magn. Reson. Imag-
ing 2000;12:965974. 2000 Wiley-Liss, Inc.
Index terms: breast cancer; MR imaging; differential diagno-
sis; contrast enhancement kinetics; BRCA; angiogenesis
UNDOUBTEDLY, MRI IS THE MOST SENSITIVE TECH-
NIQUE that is currently available for imaging primary or
recurrent breast cancer. It has been shown to be extraor-
dinarily useful for predicting disease extent, differentiat-
ing scar from recurrent cancer, identifying primary can-
cer in young high-risk patients, and evaluating tumor
response to neoadjuvant chemotherapy (15).
Despite its obvious and well-established utility, the
technique still awaits its introduction into routine clin-
ical breast imaging. There are several reasons for this,
but probably the most important one is the lack of
standardizationboth in terms of technique as well as
in terms of interpretation guidelines. If one searches the
published literature for breast MRI, one will be over-
whelmed by a myriad of technical approaches and a
seemingly unlimited number of diagnostic criteria. Ev-
ery group seems to use different techniques, with dif-
ferent criteria; none of them are concordant, everybody
chooses something else for threshold, nothing is
surein short: it is a diagnostic chaos. The conse-
quence is that potential users of breast MRI are left with
the impression that this technique is unlikely to be-
come clinically useful in the foreseeable future, because
nothing can be regarded as established knowledge.
The purpose of this review is to explain how the var-
ious techniques of dynamic breast MRI evolved, what
the underlying assumptions are, where the difculties
are, how it should be used in a clinical setting, and
most importantlyto explain what the overall tendency
in the eld is.
HISTORICAL OVERVIEW
After the introduction of gadolinium dimeglumine as
MR contrast agent, several different approaches have
been developed for MRI of the breast. Heywang et al (6)
were the rst to use gadolinium dimeglumine for MRI of
the breast. They reported strong contrast enhancement
of breast cancers, whereas the normal parenchyma ex-
hibited only weak (if any) enhancement. Heywang sug-
gested a technique that today would be called a semi-
dynamic acquisition: they acquired one pre-contrast
and two post-contrast image stacks. The main reason
for obtaining the second post-contrast stack was to
ensure detection of lesions with delayed enhancement
that may be missed on the rst post-contrast image.
Imaging was performed with limited temporal and rel-
atively high spatial resolution. Since temporal resolu-
tion was not the main focus, a 3D gradient echo tech-
nique could be applied.
The approach launched by Kaiser et al (7) was de-
signed to track the rapid signal intensity changes that
occur in the early post-contrast period. The technique
they proposed could be called the archetype of dynamic
breast MRI: they suggested acquiring one pre-contrast
and a series of post-contrast image stacks including
both breasts at the highest possible temporal resolution
(60 sec). Rapid imaging at that time allowed only a
limited spatial resolution and acquisition of only a
small number of sections (510), such that only half of
the parenchymal volume was covered. Because rapid
imaging was necessary, image subtraction was used to
suppress the signal from fatty tissues, rather than ap-
plying time-consuming active fat suppression tech-
niques.
The concept of Harms et al (8) was based on the
well-established fact that malignant lesions exhibit
characteristic morphologic features that distinguish
themfrombenign lesions. To improve analysis of subtle
morphologic details, they advocated a technique that
may serve as the archetype of static breast imaging:
imaging of one single breast with high spatial resolution
before and after contrast material injection. Since tem-
poral resolution was not an issue in this approach, 3D
gradient echo imaging was used, and fat suppression
ensued by means of spectral pre-pulses (which were
rather time consuming).
Department of Radiology, University of Bonn, Bonn, Germany.
*Address reprint requests to: C.K., Department of Radiology, University
of Bonn, Sigmund-Freud-Str. 25, D-53105 Bonn, Germany. E-mail:
kuhl@uni-bonn.de
Received August 16, 2000; Accepted September 15, 2000.
JOURNAL OF MAGNETIC RESONANCE IMAGING 12:965974 (2000)
2000 Wiley-Liss, Inc. 965
The two fundamental schools that evolved (and that
were also separated geographically) were the dynamic
school and the static school. The dynamic school
(most popular in European countries) attempted to dis-
tinguish benign and malignant lesions by enhancement
characteristics at high temporal resolution imaging.
The static school (most popular in the U.S.) attempted
the same by evaluating morphologic features of en-
hancing lesions at high spatial resolution. Due to the
severe technical constraints, particularly during the
early days of breast MRI, it was necessary to choose
between either temporal or spatial resolution, depend-
ing on the diagnostic criterion that was given priority.
Accordingly, the fundamentally different approaches
published in the literature are merely a reection of the
fact that breast MRI is technically extremely challeng-
ing. The diverging demands of an optimal temporal and
spatial resolution for the detection and classication of
enhancing lesions can hardly be met even with todays
equipment. Because researchers had to cope with the
technical shortcomings of their equipment, they started
doing breast MRI at the two ends of the spectrum of
imaging techniques that are suitable for breast MRI.
It is important not to misunderstand these different
approaches as being contradictory or as being compet-
itors for the ultimate truth. They are not meant to be
used as alternatives, but have to be understood within
the clinical and technical context of the time when they
were written and published. Today, owing to the tech-
nical progress that has been made, it is possible to
integrate these demands rather than compromising on
one or the other. Therefore, modern concepts of breast
MRI strive to consider both lesion morphology and con-
trast enhancement kinetics (9,10). As a consequence,
today there is considerable agreement in terms of what
constitutes an appropriate pulse sequence for breast
MRI. It is widely accepted that temporal resolution is a
necessitynot only if one wishes to evaluate contrast
enhancement kinetics, but also to improve the analysis
of morphological details (1012). This is due to the fact
that lesion-to-parenchyma contrast is best only in the
early post-contrast period, whereas it deteriorates pro-
gressively in the intermediate and late post-contrast
phase.
PATHOPHYSIOLOGICAL BASIS
The pathophysiological basis of lesion contrast en-
hancement in breast MRI has not yet been fully eluci-
dated, but some fundamental facts are known that
should help understand the techniques specic
strengths and weaknesses in terms of lesion detection
and differential diagnosis.
It is a well established fact that malignant lesions
release angiogenic factors (e.g., vascular endothelial
growth factor (VEGF)) that induce sprouting and
growth of pre-existing capillaries, and induce the de
novo formation of new vessels (1315). As revealed by
histologic and electron microscopic studies, these cap-
illaries exhibit a pathologic vessel wall architecture with
leaky endothelial linings. Thus, the effect of angiogenic
activity is twofold: there is an increased vascularity
(vessel density), leading to a focally increased inow of
contrast material, plus an increased vessel permeabil-
ity, leading to an accelerated extravasation of contrast
material at the site of the tumor. Because the regular
capillary architecture is only poorly reconstructed, ar-
terio-venous shunts are another hallmark of tumor-
induced angiogenesis, leading to perfusion shortcuts.
To date, however, it is unclear what exactly deter-
mines the degree of contrast material enhancement
seen on the MR image. Many studies have been pub-
lished, correlating vessel density with signal intensity
changes (1625). The results are contradictory, but
what can be stated thus far is that vessel density itself
cannot be the only contributor. A possible reason for
the inconsistent correlation between MR-detected en-
hancement and vessel density or prognostic factors is
the fact that the gadolinium-induced signal intensity
increase in T1-weighted MR images is not exactly pro-
portional to the amount (or concentration) of contrast
material that accumulates in a lesion. Lesion enhance-
ment is determined by a variety of contributing factors,
including vessel permeability, but also contrast mate-
rial diffusion rates, composition of the interstitial tumor
matrix, and baseline and post-contrast tissue T1 relax-
ation times. Because signal intensity in susceptibility-
based T2*-weighted rst-pass perfusion imaging is
more directly related to vessel density and angiogene-
sis-induced pathologic vessel permeability, it has been
suggested to use this technique as an adjunct to reg-
ular T1-weighted dynamic imaging to improve differ-
ential diagnosis of enhancing lesions (26).
What is even more problematic from the clinical ra-
diologists perspective is the fact that a locally increased
vascularity and/or capillary permeability is by no
means specic for malignant tissues. Almost all benign
neoplastic lesions, and many benign non-neoplastic
states, go along with a signicant hypervascularity or
hyperemia. Accordingly, contrast enhancement per se,
or even strong and rapid contrast enhancement, is not
a feature that is reserved for malignant lesions. How-
ever, a low vascular density can be also found in some
malignant changes. Although a non-enhancing inva-
sive breast cancer is so rare that this nding merits a
case report (27), tumors with only shallow enhance-
ment do occur in up to 10% of cases, notably in true
lobular-invasive cancers and in the scirrhotic or des-
moplastic type of ductal invasive breast cancer. Based
on histochemical studies it is now assumed that the
entire process of angiogenesis differs in these types of
breast cancers. There is evidence that in lobular-inva-
sive cancers angiogenesis is mediated by angiogenic
factors other than VEGF (28). Some well-differentiated
invasive cancers (e.g., the tubular type) may go without
a signicant degree of angiogenesis as well. Moreover,
an interaction between tumor cells with the adjacent
stroma is not necessarily found in in situ cancers
(29,30). So while a certain degree of angiogenic activity
seems to be a prerequisite for tissue invasion, and is
thus closely associated with malignant growth, this is
not necessarily to be expected for DCIS. Accordingly,
contrast enhancement of DCIS can be predicted to vary
even more than that of invasive cancers and will be
below any reasonable enhancement thresholds in a
considerable number of cases.
966 Kuhl and Schild
These pathophysiologic facts explain why vascular-
ity, and hence contrast enhancement patterns, vary
such that a clear-cut differential diagnosis based on
contrast enhancement should prove impossible. It is
quite evident that contrast enhancement itself cannot
be more specic or more sensitive than the biological (or
pathophysiologic) basis it stands for: hypervascularity
(or lack thereof) is not pathognomonic for malignant or
benign lesions. Yet, suprisingly enough, the differential
diagnostic power of evaluating lesion contrast enhance-
ment is somewhat better than one might expect given
the nonspecic distribution of vessel densities among
benign and malignant lesions. The explanation for this
phenomenon is probably the fact that it is not the mere
number of vessels, but rather the entirety of vessel
architecture, permeability, and tissue relaxation times
that determines contrast enhancement, and, thus, dif-
ferential diagnosis in dynamic breast MRI.
DIAGNOSTIC CRITERIA AND THEIR ACCURACY
Based on the imaging technique proposed by Kaiser et
al (7) and Heywang et al (31), many more dynamic
acquisition schemes have been developed, and almost
as many interpretation guidelines. For analysis of le-
sion enhancement, Heywang et al (31) suggested quan-
tifying a normalized enhancement ratio by assessing
the lesions signal intensity relative to the signal of fatty
tissue. After analyzing normalized lesion signal inten-
sities (NU) in a cohort of 144 patients with benign and
malignant breast diseases, they proposed a classica-
tion scheme that was based on lesion peak signal in-
tensity increase. Lesions with a maximum signal inten-
sity increase at or beyond the threshold of 300 NU were
classied as malignant. Between 250300 NU they
were rated borderline, and an enhancement below 250
NU was considered nonsignicant. Accordingly, their
approach to differential diagnosis would be based on
the question: How strongly does the lesion enhance?
Using this criterion, they reported a sensitivity of 100%
(71/71), and specicity of 27% (20/73).
Kaiser et al (7) suggested a quantication of lesion
enhancement as well. However, they proposed normal-
izing enhancement not with respect to the signal from
fatty tissue, but with respect to baseline lesion signal
intensity, according to the equation:
[(SI
post
SI
pre
)/SI
pre
] 100,
where SI
pre
is the signal intensity before contrast, SI
post
is the signal intensity after contrast administration. In
a preliminary study of 25 patients, they found that
breast cancers exhibit faster enhancement rates, caus-
ing a strong signal increase in the early post-contrast
period. For differential diagnosis, enhancement velocity
(relative SI increase per minute in the early post-con-
trast period) was suggested. Kaiser found that malig-
nant lesions revealed an enhancement velocity beyond
a threshold of 100% (i.e., doubling signal intensity
within the rst post-contrast minute). Accordingly, they
suggested establishing the differential diagnosis based
on the question: How fast does the lesion enhance?
Using this approach in a preliminary series of 25
patients, Kaiser reported a sensitivity of 100% (6/6).
Unfortunately, however, the authors failed to validate
their data in a larger series of patients. In several sub-
sequent review articles that were published in the rst
half of the last decade, Kaiser and coworkers stated that
the sensitivity and specicity of their approach was
99% and 98%, respectively (32,33). Notwithstanding
the rather poor validation, this imaging technique and
its associated criteria have gained considerable popu-
larity. Stomper et al (34), using the same technique on
a small series of patients, reported a sensitivity of 92%
(23/25), but achieved only a moderate specicity (61%,
16/26).
As did Kaiser, Gilles et al (35) suggested a dynamic
technique focussing on temporal resolution. They
found that differential diagnosis could be based on de-
termining the time point of lesion enhancement relative
to arterial enhancement, and they suggested that every
lesion that exhibits enhancement on the rst post-con-
trast image (i.e., 94 sec post-injections) was to be con-
sidered malignant. Accordingly, their approach to dif-
ferential diagnosis would be based on the question:
When does the lesion start to enhance?
Gilles and coworkers tested their approach in a series
of 134 patients with 64 malignant and 79 benign le-
sions. They achieved a sensitivity of 95% (61/64) and a
specicity of 53% (42/79).
The concept of analyzing onset of lesion enhancement
was also pursued by Boetes et al (36). They used an
ultrafast imaging technique that sacriced coverage of
the entire breast parenchyma in favor of a single-sec-
tion, ultra-fast bilateral image acquisition based on
turbo gradient echo sequences with a temporal resolu-
tion of 2.3 sec. Boetes and colleagues observed that
malignant lesions start to enhance 11.5 sec after bolus
arrival in the descending aorta. Moreover, they reported
that the spatial distribution of contrast material en-
hancement within a lesion differs for breast cancers
and benign lesions. They observed that in malignant
lesions enhancement starts in the periphery and
progresses from there in a centripetal fashion. Benign
tumors, on the other hand, showa centrifugal enhance-
ment pattern. Accordingly, their approach to differen-
tial diagnosis would be based on the question: When
does the lesion start to enhance, and where within the
lesion does it start?
Using these criteria in a cohort of 87 lesions, Boetes
et al (36) achieved a sensitivity and specicity of 95%
(62/65) and 86%(19/22), respectively. Schorn et al (37)
tried the same technique and the same diagnostic cri-
teria on a series of 35 lesions (15 malignant and 20
benign). In their small series, they could not reproduce
a statistically signicant difference concerning the pro-
gression of enhancement (centripetal or centrifugal) or
the onset of lesion enhancement for benign and malig-
nant lesions. The reason for this discrepancy remains
speculative; but differences concerning the distribution
of benign and malignant lesions in the respective study
groups or concerning the average lesion size may ac-
count for it.
Fischer et al (38) as well as our group (10) suggested
that in addition to the early post-contrast period, the
Dynamic Image Interpretation of Breast MRI 967
Figure 1.
968 Kuhl and Schild
intermediate and late post-contrast phases also yield
diagnostically useful information. Analysis of lesion
contrast enhancement behavior in these phases can be
used as an additional criterion with analyses of other
dynamic imaging features. We suggested a qualitative
evaluation of lesion signal intensity time courses based
on a visual classication of the shape of the time/signal
intensity curve. The classication scheme distin-
guishes types 1a and 1b, type 2, and type 3 time
courses. Enhancement is classied as type 1a if the
lesion continues to enhance over the entire acquisition
period. It is classied as type 1b if in the late post-
contrast phase the signal gain is slowed down, yielding
a bowing of the signal curve. Enhancement is classied
as type 2 if the signal plateaus after the early increase.
A type 3 curve is assigned in cases where there is a loss
of signal intensity due to wash-out of contrast material
occurring immediately after the signal intensity peak.
Lesions with steady signal intensity increase (types 1a
and 1b) were more likely to be benign, whereas lesions
with signal intensity plateau (type 2) or with a wash-out
of contrast material (type 3) tend to be malignant. Ac-
cordingly, this approach to differential diagnosis would
be based on the question: What happens after the ini-
tial signal increase?
Using this criterion, in a cohort of 266 contrast-en-
hancing lesions (10) qualitative evaluation of signal in-
tensity time courses yielded a sensitivity of 91% (92/
101) and a specicity of 83% (137/164).
All these approaches were triggered by the fact that
while breast cancers exhibit contrast enhancement,
many benign lesions do so as well. Consequently, for
differential diagnosis it is necessary to assess addi-
tional lesion features. The many different ways to as-
sess contrast enhancement kinetics, and the various
terms describing kinetic parameters that have been
listed, should not be misunderstood as discrepancies or
scientic inconsistency. Rather, they represent differ-
ent ways to look at the same phenomenon: the early,
rapid, and strong signal intensity increase that occurs
in breast cancers. Likewise, the many different thresh-
olds that have been proposed to establish cut-off val-
ues for suspicious enhancement should not be mis-
understood as giving proof of an inherent inconsistency
in the dynamic approach. The exact numbers of thresh-
old values will vary greatly with the eld strength, pulse
sequence, and timing of contrast material injection. As
such, they are not stand-alone data, and they are not
valid except for the particular setting in which they
have been established.
TECHNICAL ISSUES
The use of a dedicated surface coil is a prerequisite for
breast MRI, be it static or dynamic. Usually, for dy-
namic imaging a double-breast coil is used to allow
imaging of both breasts simultaneously. Owing to the
diverging demands of an adequate temporal and an
optimum spatial resolution, breast MRI is technically
demanding and clearly prots from high magnetic
elds. It has been shown, however, that breast MRI it
can be successfully performed at mid-eld systems
(0.5T) (39,40).
The basis of dynamic breast MRI is the T1-weighted
gradient echo pulse sequence. Due to their superior
T1-contrast and shorter acquisition times, gradient
echo sequences are generally preferred over spin echo
pulse sequences. For dynamic imaging, both 2Dand 3D
acquisition schemes are suitable. The exact design of
the dynamic series will vary depending on the diagnos-
tic criterion that is primarily used: If onset of enhance-
ment is to be evaluated, then ultrafast imaging is nec-
essary. With current state-of-the-art equipment this
will only be possible with a single-section technique.
Therefore, this approach will only be suitable for lesion
characterization (not for lesion detection) because the
location of the lesion must be known in advance to
position the slice. Accordingly, this technique has not
Figure 1. Preoperative breast MRI for staging of suspected breast cancer. A 57-year-old patient received a screening mammo-
gram. A solitary spiculated mass in the lower-outer quadrant of her right breast was rated as BIRADS 5. Breast MRI was
performed preoperatively because breast conservation was considered. Dynamic breast MRI was performed using our standard
T1-weighted 2D gradient echo series (TR/TE/FA 260/4.6/90). One image stack (with 33 sections) was acquired before contrast,
and ve were obtained after bolus injection of 0.1 mmol/kg BW gadolinium dimeglumine. Acquisition time of each image stack
(temporal resolution) 1 min 45 sec; imaging matrix 390 512, section thickness 3 mm. Image subtraction was used to
suppress the signal from fatty tissue. Signal intensity time courses were calculated. a: pre-contrast image of the known lesion
in the lower outer quadrant; b: rst post-contrast image of the dynamic series of the same location; c: subtracted image;
d: pre-contrast image of a section through the cephalad parts of the upper quadrant; e: rst post-contrast image of the dynamic
series of the same location; f: subtracted image (ef); g: signal intensity time course of the lesions in the upper quadrant;
h: maximum intensity projection image of all sections of the rst post-contrast dynamic image stack (subtracted). Note the
stellate lesion in the lower outer quadrant, with rapid and strong signal intensity increase in ac. The lesion corresponds to the
mammographically visible index lesion. Note the irregular conguration and the heterogeneous internal architecture that is
clearly visualized. In addition to the mammographically visible lesion in a, two other enhancing lesions were identied in the
upper inner and upper outer quadrant (dg). Note the rapid enhancement, the irregular morphology, and the suspicious contrast
enhancement pattern with wash-out of the signal intensity time course (curve type 3). The lesions were rated as highly
suspicious for invasive breast cancer. The MIP image (h) provides a good overviewon the location of the lesions in the lower outer,
upper outer, and upper inner quadrant. Breast MR conrms the absence of breast cancer in the left breast. The patient was
operated on after MR-guided localization of the clinically and conventionally occult lesions in the upper part of the breast.
Histology revealed a multicentric but highly differentiated invasive tubular carcinoma (one in the lower quadrant, 6 mm in size;
and two in the upper quadrants, 4 mm each), yielding a stage pT1b
m
N0M0. Owing to the small size of the lesions relative to the
size of the breast, and due to the good prognosis of tubular cancer, breast conservation surgery was performed.
Dynamic Image Interpretation of Breast MRI 969
gained widespread use. The vast majority of dynamic
imaging techniques that are used for clinical breast
imaging today are designed to allow evaluation of ki-
netic and morphologic parameters (4,1012,41). There-
fore, an in-plane pixel size of about 1 mm is recom-
mended, with a section thickness of no more than 3
mm. To be able to track the rapid signal intensity
changes that occur in the early post-contrast period,
and to ensure high lesion-to-parenchyma contrast, a
minimumtemporal resolution is required. What exactly
constitutes minimum temporal resolution is, how-
ever, still a matter of debate. It is generally felt that the
acquisition time of one dynamic image stack should
take less than two minutes; ideally, it should be around
one minute. (These values are derived from clinical
practice; as yet, these recommendations have not been
substantiated by prospective trials.) Because bilateral
imaging is performed over a relatively large eld of view
(FOV), and because a rapid image acquisition is neces-
sary, active fat suppression techniques are not an op-
tion. Instead, image subtraction is used to suppress the
signal from fatty tissue.
The dynamic imaging protocol used at our institution
may serve as an example (Fig. 1). It is a 2D gradient
echo series with TR/TE/FA 260/4.6/90; 390 512
imaging matrix; FOV of 290310 mm; section thickness
of 3 mm, with 33 sections; and temporal resolution of
1:45 sec per dynamic stack. One set of images is ac-
quired before contrast, and another ve image stacks
after bolus i.v. injection (4 mL/sec) of 0.1 mmol/kg
gadolinium dimeglumine, followed by a saline ush. To
quantify enhancement, signal intensity is measured via
ROIs that are selectively placed into the area of a lesion
in which the earliest enhancement occurs. Enhance-
ment velocity is quantied via the enhancement for-
mula mentioned above; the time course of signal inten-
sity is assessed visually.
HOW TO USE INFORMATION OBTAINED FROM
DYNAMIC DATA
Owing to the biologic and histologic heterogeneity of
breast lesions mentioned above, it should go without
saying that it is not possible to diagnose breast cancers
with a simple enhancement cut-off value or threshold.
An enhancement threshold may enable the character-
ization of a substantial number of enhancing le-
sionsas long as they exhibit a typical contrast en-
hancement. Yet, because far from all lesions enhance
as expected, this concept will cause an unacceptably
high rate of false-positive or (even more fatal) false-
negative decisions on lesions that behave less typically.
To avoid any false-negative decisions, a threshold must
be set at relatively moderate enhancement values; how-
ever, a myriad of benign lesions will then reach supra-
threshold enhancement. There is extensive evidence in
the literature suggesting that many benign lesions (-
broadenomas, focal adenosis, proliferative dyplasia,
papillomas, focal chronic mastitis, fresh fat necrosis,
hyperplastic intramammary lymph nodes, and even
normal breast parenchyma under hormonal stimula-
tion) may go along with contrast enhancement rates
that may be well beyond any reasonable cut-off value
(26,4244). On the other hand, even with low cut-off
values, there will be breast cancers that fail to meet the
required enhancementmostly lobular, scirrhotic
ductal, mucinous, and tubular invasive cancers are
candidates for such below-threshold enhancement
(45). As a consequence, dynamic data (enhancement
velocity, degree, onset, and so forth) must not be used
as stand-alone diagnostic criterion. Instead, they
should be integrated in the process of differential diag-
nosticto expand (rather than narrow) the armamen-
tarium of differential diagnosis in breast MRI.
We use the following guidelines (10): For interpreta-
tion of dynamic breast MR images, we rst refer to the
rst post-contrast image stack and search for lesions
with signicant enhancement, because lesions that ap-
pear at this phase are associated with a signicant
probability of malignancy. Once such a lesion is iden-
tied, morphology is evaluated. If morphology is suspi-
cious, biopsy is recommended. If morphology is equiv-
ocal or benign, the time course of signal intensity is
evaluated. If a type 3 time course is identied, biopsy is
recommended. If it is a type 1 or 2 time course, we
usually recommend follow-up.
If a lesion with shallow enhancement is identied, the
management decisions are based solely on lesion mor-
phology, in order to include, e.g., lobular cancer or
DCIS (46). The lesions internal architecture (47) is con-
sidered such that if rim enhancement is identied bi-
opsy is recommended irrespective of other ndings. If
internal septations are clearly discernible, diagnosis of
broadenoma is establishedagain irrespective of
other ndings, including kinetic data. In addition, as
adjunctive criteria we consider the progression of en-
hancement (centripetal/centrifugal), and the lesions
signal intensity in T2-weighted images (48).
It should be well understood that these guidelines are
not carved in stone, but must be adapted to the indi-
vidual patients situation by considering the clinical
ndings, the ndings in conventional imaging modali-
ties, the patients age, her medical history including
menstrual status or hormone medication, and family
history.
CURRENT APPLICATIONS OF DYNAMIC
BREAST MRI
There are several indications for breast MRI; however,
at this stage, there are virtually no data available that
would allow end-point analyses to be performed on the
inuence of breast MRI in terms of survival, mortality,
morbidity, or quality of life issues.
Clarication of Inconclusive Conventional
Imaging Findings
Breast MRI offers a wealth of information on the lesion
in question. It provides detailed information on high-
resolution, high-contrast cross-sectional tumor mor-
phology, and insight into tumor biology, angiogenetic
activity, T1- and T2-relaxation rates, contrast agent
relaxivity, etc.all the physical and biochemical fea-
tures that determine image contrast on MR studies. The
many parameters that contribute to lesion appearance
970 Kuhl and Schild
translate into a full battery of differential diagnostic
criteria. These may be used to distinguish benign and
malignant lesions even in cases that are inconclusive
on conventional imaging. Specically, dynamic breast
MRI may be useful in evaluating lesions that appear
morphologically benign on conventional imaging stud-
ies. The evaluation of time course kinetics introduces a
completely independent diagnostic parameter (i.e., tis-
sue perfusion/diffusion/vessel permeability) that can
help distinguish benign lesions (e.g., broadenoma)
from well-circumscribed breast cancer. If, for example,
a breast cancer looks benign in terms of morphology, a
correct diagnosis may still be possible if signal intensity
time courses are evaluated (4,10,11).
It should be noted, however, that if breast MRI is to be
used for clarication of mammographically or sono-
graphically suspicious lesions, then it is an important
prerequisite that the radiologist be familiar with the
specic limitations of all three imaging modalities. It is
important to realize that there are specic constella-
tions of mammographic or sonographic ndings that
may not be claried by a negative breast MRI, whereas
in others MRI can be used to obviate the need for bi-
opsy. The former holds true for example, for cases with
suspicious mammographic microcalcications. Be-
cause sensitivity of breast MRI for in-situ cancers is
limited, it may not be used to exclude underlying DCIS
(30). The latter holds true for example, if tumor recur-
rence has to be ruled out in a stellate density after
breast conservation therapy (49). Moreover, it should
be remembered that, in general, percutaneous core bi-
opsy may be more appropriate to denitively clarify
conventionally inconclusive lesions.
Staging
If on conventional imaging studies a solitary focus of
breast cancer has been identied and a breast conserv-
ing therapy is considered, preoperative breast MRI is
indicated to rule out or localize additional breast cancer
foci. In a recent article, Fischer et al (41) reported on
preoperative dynamic breast MRI for local staging of
patients who were candidates for breast conservation.
They diagnosed therapeutically relevant additional
ndings in 16% of cases. Because dynamic breast MRI
(as opposed to high-spatial-resolution static breast
MRI) allows the simultaneous evaluation of both
breasts, a screening of the contralateral breast will
always be performed. This is reasonable because a syn-
chronous contralateral breast cancer will be present in
as many as 6% of patients (41).
Assessing Tumor Response to Neoadjuvant
Chemotherapy
Neoadjuvant chemotherapy is increasingly used in pa-
tients with locally advanced breast cancer (LABC) for
restoring operability as well as for systemic treatment of
possible concomitant lymph node or distant metasta-
ses. Conventional imaging techniques, however, offer
only a poor diagnostic accuracy for the assessment of
chemotherapeutic effects (50). This is mainly due to the
fact that after effective chemotherapy tumor tissue may
be replaced by diffuse brosis. The brous tissue may
simulate residual tumor upon clinical palpation, and it
may interfere with an accurate depiction of residual
tumor on both mammograms and breast ultrasound
(US) studies. Moreover, for optimizing patient care as
well as for economic reasons, it is crucial to reliably
identify non-responders as soon as possible. There is
evidence that dynamic breast MRI is ideally suited to
fulll both of these tasks.
Evaluation of Chemotherapy Response
In addition to tumor morphology, dynamic breast MRI
is able to quantify functional tissue parameters, such
as tumor perfusion, as a surrogate marker for tissue
viability. Because dynamic breast MRI is able to detect
and quantify chemotherapy-induced changes of malig-
nant tissues perfusion or viability, this can be exploited
for assessing tumor response to neoadjuvant chemo-
therapy. Several groups have investigated whether dy-
namic breast MRI can identify responders or non-re-
sponders (5,5154). With a remarkable consistency, the
different studies revealed that after just one or two
chemotherapy cycles, and before a measurable change
of tumor size occurred, response to chemotherapy was
heralded by a substantial change of contrast enhance-
ment patterns. Decreasing enhancement rates, a at-
tening of the signal intensity time course, and a re-
duced degree of enhancement are the hallmarks of
early tumor response to chemotherapy. Although the
sample size of the studies is small, there is sufcient
evidence to conclude that patients with a completely
unchanged contrast enhancement pattern after two
chemotherapy cycles can be classied as non-respond-
ers. Given the high clinical and economical relevance of
an early identication of non-responders, this may
emerge as one of the most important applications for
dynamic breast MRI.
Evaluation of Residual Tumor
Several studies have conrmed thatcompared to con-
ventional imaging modalitiesbreast MRI is much
more sensitive and specic for assessing residual tu-
mor extent after chemotherapy (51,5557). It should be
well understood, however, that although breast MRI
may be better than clinical assessment or conventional
imaging, it is still far from being perfect. Scattered re-
sidual vital cancer cells in the former tumor bed in
responders may not enhance after contrast, thus es-
caping the diagnosis. It is a well established fact that
these tumor remnants do not inuence prognosis and,
thus, the classication of response; however, if induc-
tion chemotherapy is performed with the ultimate goal
of breast conservation, breast MRI cannot be used to
rule out residual micro-manifestations in responders,
and it may not be able to identify patients who are
amenable to breast conservation after induction che-
motherapy.
High-Risk Screening
Women with proved BRCA mutation or with a family
history suggestive of hereditary breast cancer face an
Dynamic Image Interpretation of Breast MRI 971
8090% lifetime risk of being diagnosed with breast
cancer. Moreover, these women tend to develop breast
cancer at signicantly younger ages (i.e., in their early
thirties) than the genetically intact woman. Accord-
ingly, these women require an intensied screening
starting at age 2530. In these very young women, how-
ever, the sensitivity of mammography may be signi-
cantly reduced. Recently, we published our experiences
with dynamic breast MRI screening in high-risk women
diagnosed or suspected to carry a breast cancer sus-
ceptibility gene (4). Our results document that dynamic
breast MRI is clearly superior to both mammography
and breast US for early detection and classication of
breast cancers. The dynamic approach seems to be
particularly useful here for three reasons:
First, in a screening setting, bilateral imaging is re-
quired. Dynamic imaging is almost always done with
coverage of both breasts, whereas high-spatial-resolu-
tion static MRI can only be performed on one breast at
a time.
Second, our results showthat BRCA1-induced breast
cancers in particular exhibit atypical morphologic fea-
tures. Even on gross pathology or low-magnication
histology, these tumors may appear completely well-
circumscribed, with no evidence of inltration. Accord-
ingly, even with the highest spatial resolution imaging,
these tumors may not be distinguishable from the
many broadenomas that plague mammographic inter-
pretation of patients of this age group. Dynamic breast
MRI, however, allows additional evaluation of tissue
perfusion or angiogenetic activity. We have shown that
tumors with deceptively benign morphology exhibit
highly suspicious contrast enhancement kinetics with
early, rapid, and strong signal intensity increase, fol-
lowed by a wash-out of signal intensity (time course
type 3). Thus, in spite of the apparently benign morpho-
logic features of hereditary breast cancers, a true-pos-
itive diagnosis was possible in all cases based on dy-
namic MRI. Accordingly, sensitivity of breast MRI vs.
mammography and breast US combined was 100% vs.
44%.
Third, in very young women, hormonal stimulation
may produce pseudo-lesions, i.e., focal contrast-en-
hancing areas that are not associated with any struc-
tural changes of the parenchymal composition but are
probably caused by the local histamine-like effects of
ovarian steroid hormones. These pseudo-lesions are ex-
tremely prevalent in younger patients; they are a noto-
rious cause of false-positive ndings in breast MRI
studies. It has been shown that these lesions can ex-
hibit an alarmingly irregular morphology. However,
contrast enhancement kinetics (particularly the signal
intensity time courses) correspond to the benign type
1 time course in the vast majority of cases. Thus time
course analysis can help correctly classify these pseu-
do-lesions as benign. Accordingly, in our series of al-
most 200 women receiving 350 MRI studies, breast MRI
had the lowest false-positive rate of all imaging modal-
ities under investigation (mammography, US, and MRI).
Therefore, the positive predictive value of breast MRI
(64%) was signicantly higher than that of mammogra-
phy (44%) or breast US (12%).
It can be concluded that dynamic breast MRI enables
adequate surveillance and early diagnosis of the high-
risk patient with hereditary breast cancer. Accordingly,
breast MRI screening may someday be used as an al-
ternative to prophylactic bilateral mastectomy, which is
currently the standard treatment option for the many
young women who are suspected gene carriers. Further
studies (particularly multicenter outcome studies) will
have to elucidate whether MRI screening can effectively
replace bilateral prophylactic mastectomy for prevent-
ing BRCA-induced breast cancer mortality.
Assessing Tumor Grade and Prognosis
It has been extensively shown that tumor vascularity,
as revealed by histologic vessel density counts, corre-
lates with tumor aggressiveness and malignant (partic-
ularly metastatic) potential (29,5861,63). Because
contrast enhancement patterns in dynamic breast MRI
seemto be linked with hypervascularity (1618,2022),
the intriguing thought came up that dynamic breast
MRI might be useful for tumor grading, or assessing
tumor aggressiveness or prognosis in vivo. Several re-
ports have been published investigating the correlation
between contrast enhancement rates or time course
features and prognostic factors such as histological
grading, lymph node status, S-phase fraction, and
modern proliferation indices (oncogenes/tumor sup-
pressor genes) such as c-erbB-1, c-erB-2, p53, or Ki-67.
Unfortunately, however, the results of these studies
were inconsistent. In two prospective studies, no corre-
lation with any of the criteria used in dynamic breast
MRI (enhancement rates, maximum enhancement,
wash-out rates, etc.) has been obtained (62,64). Yet
there are two studies that revealed a highly signicant
correlation between contrast enhancement in dynamic
MRI and prognostic factors: Mussurakis (65) reported
on a strong correlation of enhancement rates and tu-
mor grading as well as nodal status in 53 patients with
invasive breast cancers; Bone et al (66) conrmed these
ndings in another 50 breast cancer cases.
To date the source of the discrepancy between the
published studies is unclear. It is possible that the
variability in determining contrast enhancement rates
if manually drawn ROIs are used may account for some
heterogeneity of results. This is supported by the report
of Mussurakis et al (67) that a statistically signicant
correlation between enhancement rates and prognostic
factors was only obtained if an automatic ROI denition
based on parametric images was used (51).
Studies on contrast-enhanced dynamic breast MRI of
R3230 implanted adenocarcinoma have been per-
formed using new blood-pool contrast agents, allowing
the assessment of tumor vessel permeability, and thus
of tumor angiogenic activity grading (6871). Hopes are
high that once these agents are available for use in
humans, a noninvasive in vivo grading of breast can-
cers based on preoperative breast MRI will become fea-
sible.
FUTURE DIRECTION
In the eld of breast MRI, extensive discussions over
discrepant technologies and interpretation guidelines
972 Kuhl and Schild
have for a long time prevented the acceptance of the
technique, and precluded the organization and perfor-
mance of multi-institutional trials that are urgently
needed to document the long-term utility of the tech-
nique in breast cancer patients. The most important
argument against the introduction of breast MRI into
routine clinical practice is that it is a nonstandardized
and nonstandardizable technique, with poor reproduc-
ibility. This attitude, however, ignores the fact that the
seemingly chaotic variety of technical and interpreta-
tional approaches is in fact a direct consequence of the
techniques major advantage: Breast MRI, compared to
mammography, for example, yields a wealth of informa-
tion on the tissue under investigation, including cross-
sectional morphology as well as tissue relaxation times,
perfusion, and diffusion as revealed by contrast en-
hancement kinetics. Yet, understandably, and as a di-
rect consequence, it will take longer to reach consensus
on what is diagnostically relevant if one has about
seven different criteria to work outas opposed to, e.g.,
the situation in mammography, in which merely two
criteria (morphology and x-ray density) contribute to
the diagnosis. Today, the breast MRI community is nav-
igating toward a consensus in terms of indications,
techniques, diagnostic criteria, and overall appraisal of
the techniques clinical use. There is broad agreement
that information on contrast enhancement kinetics as
provided by dynamic breast MRI should not be used as
stand-alone diagnostic criteria, but must be inte-
grated into the process of lesion differential diagnosis to
improve the early detection and correct classication of
breast cancer.
REFERENCES
1. Heywang-Kobrunner SH, Viehweg P, Heinig A, Kuchler C. Contrast
enhanced MRI of the breast: accuracy, value, controversies, solu-
tions. Eur J Radiol 1997;24:94108.
2. Orel SG, Schnall MD, Powell M, Hochmann MG, Solin LJ, Fowble
BL, Torosian MH, Rosario EF. Staging of suspected breast cancer:
effect of MR imaging and MR-guided biopsy. Radiology 1995:196:
115122.
3. Gilles R, Guinebretiere JM, Shapeero LG, Lesnik A, Contesso G,
Sarrazin D, Masselot J, Vanel D. Assessment of breast cancer
recurrence with contrast-enhanced subtraction MR imaging: pre-
liminary results in 26 patients. Radiology 1993;188:473478.
4. Kuhl CK, Schmutzler RK, Leutner CC, Kempe A, Wardelmann E,
Hocke A, Maringa M, Pfeifer U, Krebs D, Schild HH. Breast MR
imaging screening in 192 women proved or suspected to be carriers
of a breast cancer susceptibility gene: preliminary results. Radiol-
ogy 2000;215:267279.
5. Kurtz B, Achten C, Audretsch W, Rezai M, Urban P, Zocholl G.
MR-mammography assessment of tumor response after neoadju-
vant radiochemotherapy of locally advanced breast carcinoma. MR-
mammographische Beurteilung des Tumoransprechverhaltens
nach neoadjuvanter Radiochemotherapie lokal fortgeschrittener
Mammakarzinome. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb
Verfahr 1996;164:469474.
6. Heywang SH, Hahn D, Schmidt H, Krischke I, Eiermann W, Basser-
mann R, Lissner J. MR imaging of the breast using gadolinium-
DTPA. J Comput Assist Tomogr 1986;10:199204.
7. Kaiser WA, Zeitler E. MR imaging of the breast: fast imaging se-
quences with and without Gd-DTPA. Preliminary observations. Ra-
diology 1989;170:681686.
8. Harms SE, Flamig DP, Hesley KL, Meiches MD, Jensen RA, Evans
WP, Savino DA, Wells RV. MR imaging of the breast with rotating
delivery of excitation off resonance: clinical experience with patho-
logic correlation. Radiology 1993;187:493501.
9. Schnall MD, Ikeda DM. Lesion diagnosis working group report. J
Magn Reson Imaging 1999;10:982990.
10. Kuhl CK, Mielcarek P, Klaschik S, Leutner C, Pakos E, Gieseke J,
Schild H. Are signal time course data useful for differential diag-
nosis of enhancing lesions in dynamic breast MR imaging? Radiol-
ogy 1999;211:101110.
11. Liu PF, Debatin JF, Caduff RF, Kacl G, Garzoli E, Krestin GP.
Improved diagnostic accuracy in dynamic contrast enhanced MRI
of the breast by combined quantitative and qualitative analysis.
Br J Radiol 1998;71:501509.
12. Kinkel K, Helbich TH, Esserman LJ, Barclay J, Schwerin EH, Sick-
les EA, Hylton NM. Dynamic high-spatial-resolution MR imaging of
suspicious breast lesions: diagnostic criteria and interobserver
variability. Am J Roentgenol 2000;175:3543.
13. Folkman J, Klagsbrun M. Angiogenic factors. Science 1987;235:
442447.
14. Folkman J, Watson K, Ingbr D, Hanahan D. Induction of angiogen-
esis during the transition from hyperplasia to neoplasia. Nature
1989:339:5861.
15. Pham CD, Roberts TP, van Bruggen N, Melnyk O, Mann J, Ferrara
N, Cohen RL, Brasch RC. Magnetic resonance imaging detects
suppression of tumor vascular permeability after administration of
antibody to vascular endothelial growth factor. Cancer Invest 1998;
16:225230.
16. Buadu LD, Murakami J, Murayama S, Hashiguchi N, Sakai S,
Masuda K, Toyoshima S, Kuroki S, Ohno S. Breast lesions: corre-
lation of contrast medium enhancement patterns on MR images
with histopathologic ndings and tumor angiogenesis. Radiology
1996;200:639649.
17. Buckley DL, Drew PJ, Mussurakis S, Monson JR, Horsman A.
Microvessel density of invasive breast cancer assessed by dynamic
Gd-DTPA enhanced MRI. J Magn Reson Imaging 1997;7:461464.
18. Degani H, Gusis V, Weinstein D, Fields S, Strano S. Mapping
pathophysiological features of breast tumors by MRI at high spatial
resolution. Nat Med 1997;3:780782.
19. Bone B, Wiberg MK, Parrado C, Falkmer U, Aspelin P, Gad A.
Mechanism of contrast enhancement in breast lesions at MR im-
aging. Acta Radiol 1998;39:494500.
20. Frouge C, Guinebretiere JM, Contesso G, Di Paola R, Blery M.
Correlation between contrast enhancement in dynamic magnetic
resonance imaging of the breast and tumor angiogenesis. Invest
Radiol 1994;29:10431049.
21. Lewin M, Bredow S, Sergeyev N, Marecos E, Bogdanov Jr A,
Weissleder R. In vivo assessment of vascular endothelial growth
factor-induced angiogenesis. Int J Cancer 1999;83:798802.
22. Hulka CA, Edmister WB, Smith BL, Tan L, Sgroi DC, Campbell T,
Kopans DB, Weisskoff RM. Dynamic echo-planar imaging of the
breast: experience in diagnosing breast carcinoma and correlation
with tumor angiogenesis. Radiology 1997;205:837842.
23. Buckley DL, Drew PJ, Mussurakis S, Monson JR, Horsman A.
Microvessel density of invasive breast cancer assessed by dynamic
Gd-DTPA enhanced MRI. J Magn Reson Imaging 1997;7:461446.
24. Taylor JS, Tofts PS, Port R, Evelhoch JL, Knopp M, Reddick WE,
Runge VM, Mayr N. MR imaging of tumor microcirculation: promise
for the new millennium. J Magn Reson Imaging 1999;10:903907.
25. Stomper PC, Winston JS, Herman S, Klippenstein DL, Arredondo
MA, Blumenson LE. Angiogenesis and dynamic MR imaging gado-
linium enhancement of malignant and benign breast lesions.
Breast Cancer Res Treat 1997;45:3946.
26. Kuhl CK, Bieling H, Gieseke J, Ebel T, Mielcarek P, Far F, Folkers
P, Elevelt A, Schild H. Breast neoplasms: T2* susceptibility-con-
trast, rst-pass perfusion MR imaging. Radiology 1997;202:8795.
27. Fischer U, Kopka L, Grabbe E. Invasive mucinous carcinoma of the
breast missed by contrast-enhancing MR imaging of the breast.
Eur Radiol 1996;6:929931.
28. Lee AH, Dublin EA, Bobrow LG, Poulsom R. Invasive lobular and
invasive ductal carcinoma of the breast show distinct patterns of
vasular endothelial growth factor expression and angiogenesis.
J Pathol 1998;185:394401.
29. Goede V, Fleckenstein G, Dietrich M, Osmers RG, Kuhn W, Augus-
tin HG. Prognostic value of angiogenesis in mammary tumors.
Anticancer Res 1998;18:21992202.
30. Westerhof JP, Fischer U, Moritz JD, Oestmann JW. MR imaging of
mammographically detected clustered microcalcications: is there
any value? Radiology 1998;207:675681.
31. Heywang SH, Wolf A, Pruss E, Hilbertz T, Eiermann W, Permanet-
ter W. MR imaging of the breast with Gd-DTPA: use and limitations.
Radiology 1989;171:95103.
Dynamic Image Interpretation of Breast MRI 973
32. Kaiser WA. Magnet-Resonanz-Tomographie der Brust: Ergebnisse
nach 253 Untersuchungen. Dtsch Med Wochenschr 1989;114:1351
1357.
33. Kaiser WA. MR-Mammographie. Radiologe 1993;33:292299.
34. Stomper PC, Herman S, Klippenstein DL, Winston JS, Edge SB,
Arredondo MA, Mazurchuk RV, Blumenson LE. Suspect breast
lesions: ndings at dynamic gadolinium-enhanced MR imaging
correlated with mammographic and pathologic features. Radiology
1995;197:387395.
35. Gilles R, Guinebretiere JM, Lucidarme O, Cluzel P, Janaud G, Finet
JF, Tardivon A, Masselot J, Vanel D. Nonpalpable breast tumors:
diagnosis with contrast-enhanced subtraction dynamic MR imag-
ing. Radiology 1994;191:625631.
36. Boetes C, Barentsz JO, Mus RD, van der Sluis RF, van Erning LJ,
Hendriks JH, Holland R, Ruys SH. MR characterization of suspi-
cious breast lesions with a gadolinium-enhanced TurboFLASH
subtraction technique. Radiology 1994;193:777781.
37. Schorn C, Fischer U, Luftner-Nagel S, Grabbe E. Diagnostic poten-
tial of ultrafast contrast-enhanced MRI of the breast in hypervas-
cularized lesions: are there advantages in comparison with stan-
dard dynamic MRI? J Comput Assist Tomogr 1999;23:118122.
38. Fischer U, von Heyden D, Vosshenrich R, Vieweg I, Grabbe E.
Signalverhalten maligner und benigner Lasionen in der dynamis-
chen 2D-MRT der Mamma. Rofo Fortschr Geb Rontgenstr Neven
Bildgeb verfahr 1993;158:287292.
39. Kuhl CK, Kreft BP, Hauswirth A, Gieseke J, Elevelt A, Reiser M,
Schild HH. MR-Mammographie bei 0.5Tesla: Die Differenzier-
barkeit maligner und benigner Lasionen in der MR-Mammographie
bei 0.5T und 1.5T. Rofo Fortschr Geb Rontgenstr Neven Bildgeb
verfahr 1995;162:482491.
40. Kuhl CK, Kreft BP, Hauswirth A, Elevelt A, Steudel A, Reiser M,
Schild HH. MR-Mammographie bei 0.5Tesla: Vergleich von Bild-
qualitat und Sensitivitat der MR-Mammographie bei 0.5T und
1.5T. Rofo Fortschr Geb Rontgenstr Neven Bildgeb verfahr 1995;
162:381389.
41. Fischer U, Kopka L, Grabbe E. Breast carcinoma: effect of pre-
operative contrast-enhanced MR imaging on the therapeutic ap-
proach. Radiology 1999;213:881888.
42. Kuhl CK, Kreft BP, Bieling HB, Sommer T, Lutterbey G, Gieseke J,
Schild HH. Dynamic breast MRI in premenopausal healthy volun-
teers: normal values of contrast enhancement and cycle phase
dependency. Radiology 1997;203:137144.
43. Mu ller-Schimpe M, Ohmenhauser K, Stoll P, Dietz K, Claussen
CD. Menstrual cycle and age: inuence on parenchymal contrast
medium enhancement in MR imaging of the breast. Radiology
1997;203:145149.
44. Kuhl CK. MR imaging of breast tumors. Eur Radiol 2000;10:46
58.
45. Boetes C, Strijk SP, Holland R, Barentsz JO, Van Der Sluis RF,
Ruijs JH. False-negative MR imaging of malignant breast tumors.
Eur Radiol 1997;7:12311234.
46. Kuhl CK, Mielcarek P, Leutner C, Schild HH. Diagnostic criteria of
ductal carcinoma in-situ (DCIS) in dynamic contrast-enhanced
breast MRI: comparison with invasive breast cancer (IBC) and be-
nign lesions. Proceedings of the International Society for Magnetic
Resonance in Medicine ISMRM. 1998;931.
47. Nunes LW, Schnall MD, Siegelman ES, Langlotz CP, Orel SG, Sul-
livan D, Muenz LA, Reynolds CA, Torosian MH. Diagnostic perfor-
mance characteristics of architectural features revealed by high
spatial-resolution MR imaging of the breast. Am J Roentgenol
1997;169:409415.
48. Kuhl CK, Mielcarek P, Klaschik S, Pakos E, Schild HH. Are T2-
weighted pulse sequences helpful to assist differential diagnosis of
enhancing lesions in dynamic breast MRI? J Magn Reson Imaging
1999;9:187196.
49. Heywang-Kobrunner SH, Schlegel A, Beck R, Wendt T, Kellner W,
Lommatzsch B, Untch M, Nathrath WB. Contrast-enhanced MRI of
the breast after limited surgery and radiation therapy. J Comput
Assist Tomogr 1993;17:891900.
50. Huber S, Wagner M, Zuna I, Medl M, Czembirek H, Delorme S.
Locally advanced breast carcinoma: evaluation of mammography
in the prediction of residual disease after induction chemotherapy.
Anticancer Res 2000;20:553558.
51. Rieber A, Zeitler H, Rosenthal H, Gorich J, Kreienberg R, Brambs
HJ, Tomczak R. MRI of breast cancer: inuence of chemotherapy
on sensitivity. Br J Radiol 1997;70:452458.
52. Knopp MV, Brix G, Junkermann HJ, Sinn HP. MR mammography
with pharmacokinetic mapping for monitoring of breast cancer
treatment during neoadjuvant therapy. Magn Reson Imaging Clin
N Am 1994;2:633658.
53. Knopp MV, Weiss E, Sinn HP, Mattern J, Junkermann H, Radeleff
J, Magener A, Brix G, Delorme S, Zuna I, van Kaick G. Pathophys-
iologic basis of contrast enhancement in breast tumors. J Magn
Reson Imaging 1999;10:260266.
54. Trecate G, Ceglia E, Stabile F, Tesoro-Tess JD, Mariani G, Zambetti
M, Musumeci R. Locally advanced breast cancer treated with pri-
mary chemotherapy: comparison between magnetic resonance im-
aging and pathologic evaluation of residual disease. Tumori 1999;
85:220228.
55. Gilles R, Guinebretiere JM, Toussaint C, Spielman M, Rietjens M,
Petit JY, Contesso G, Masselot J, Vanel D. Locally advanced breast
cancer: contrast-enhanced subtraction MR imaging of response to
preoperative chemotherapy. Radiology 1994;191:633638.
56. AbrahamDC, Jones RC, Jones SE, Cheek JH, Peters GN, Knox SM,
Grant MD, Hampe DW, Savino DA, Harms SE. Evaluation of neo-
adjuvant chemotherapeutic response of locally advanced breast
cancer by magnetic resonance imaging. Cancer 1996;78:91100.
57. Mumtaz H, Davidson T, Spittle M, Tobias J, Hall-Craggs MA, Cow-
ley G, Taylor I. Breast surgery after neoadjuvant treatment. Is it
necessary? Eur J Surg Oncol 1996;22:335341.
58. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angiogenesis
and metastasiscorrelation in invasive breast carcinoma. N Engl
J Med 1991;324:18.
59. Hansen S, Grabau DA, Sorensen FB, Bak M, Vach W, Rose C.
Vascular grading of angiogenesis: prognostic signicance in breast
cancer. Br J Cancer 2000;82:339347.
60. Karaiossidi H, Kouri E, Arvaniti H, Skas S, Vasilaros S. Tumor
angiogenesis in node-negative breast cancer: relationship with re-
lapse-free survival. Anticancer Res 1996;16:40014002.
61. Acenero MJ, Gallego MG, Ballesteros PA, Gonzalez JF. Vascular
density as a prognostic indicator for invasive ductal breast carci-
noma. Virchows Arch 1998;432:113117.
62. Fischer U, Kopka L, Brinck U, Korabiowska M, Schauer A, Grabbe
E. Prognostic value of contrast-enhanced MR mammography in
patients with breast cancer. Eur Radiol 1997;7:10021005.
63. Costello P, McCann A, Carney DN, Dervan PA. Prognostic signi-
cance of microvessel density in lymph node negative breast carci-
noma. Hum Pathol 1995;26:11811184.
64. Stomper PC, Herman S, Klippenstein DL, Winston JS, Budnick
RM, Stewart CC. Invasive breast carcinoma: analysis of dynamic
magnetic resonance imaging enhancement features and cell prolif-
erative activity determined by DNA S-phase percentage. Cancer
1996;77:18441849.
65. Mussurakis S, Buckley DL, Horsman A. Dynamic MR imaging of
invasive breast cancer: correlation with tumour grade and other
histological factors. Br J Radiol 1997;70:446451.
66. Bone B, Aspelin P, Bronge L, Veress B. Contrast-enhanced MR
imaging as a prognostic indicator of breast cancer. Acta Radiol
1998;39:279284.
67. Mussurakis S, Buckley DL, Coady AM, Turnbull LW, Horsman A.
Observer variability in the interpretation of contrast enhanced MRI
of the breast. Br J Radiol 1996;69:10091016.
68. Helbich TH, Gossman A, Mareski PA, Raduchel B, Roberts TP,
Shames DM, Muhler M, Turetschek K, Brasch RC. A new polysac-
charide macromolecular contrast agent for MR imaging: biodistri-
bution and imaging characteristics. J Magn Reson Imaging 2000;
11:694701.
69. Daldrup HE, Roberts TP, Muhler A, Gossmann A, Roberts HC,
Wendland M, Rosenau W, Brasch RC. Macromolecular contrast
media for MR mammography. A new approach to characterizing
breast tumors. Makromolekulare Kontrastmittel fur die MR-Mam-
mographie. Ein neuer Ansatz fur die Charakterisierung von Mam-
matumoren. Radiologe 1997;37:733740.
70. Schwickert HC, Stiskal M, Roberts TP, van Dijke CF, Mann J,
Muhler A, Shames DM, Demsar F, Disston A, Brasch RC. Contrast-
enhanced MR imaging assessment of tumor capillary permeability:
effect of irradiation on delivery of chemotherapy. Radiology 1996;
198:893898.
71. Cohen FM, Kuwatsuru R, Shames DM, Neuder M, Mann JS, Vexler
V, Rosenau W, Brasch RC. Contrast-enhanced magnetic resonance
imaging estimation of altered capillary permeability in experimen-
tal mammary carcinomas after X-irradiation. Invest Radiol 1994;
29:970977.
974 Kuhl and Schild